Curriculum Vitae
Jackson Eshbaugh — Computer Science & French, Lafayette College ’27
eshbaugj@lafayette.edu ·
jacksoneshbaugh.github.io ·
ORCID
+1 484.484.3326 · Bethlehem, PA, USA
Education
Lafayette College
B.S. Computer Science (expected 2027); A.B. French (expected 2027). GPA: 4.0/4.0.
Research Positions
Independent Research — Neural Network Interpretability
- Investigate when linear surrogate models fail to faithfully represent neural networks.
- Design experiments comparing surrogate fidelity with task accuracy across regression tasks.
- Proposed the λ-score as a diagnostic metric and implemented a full ML pipeline.
EXCEL Scholar — Building Energy & Generative AI
- Develop neural-network approaches to recommend energy-efficiency retrofits at neighborhood scale.
- Use EnergyPlus simulations and generative AI for urban building energy modeling.
- Co-author on the “Synthetic Homes” manuscript and related poster presentations.
Honors Thesis Research — Computational Linguistics
- Develop FRIdiom, an annotated corpus for detecting French idiomatic expressions.
- Apply back-translation and neural methods to figurative language in multilingual MT.
- Use interpretability techniques to probe language model representations.
ACL Manuscript Review Collaborator
- Serve as a secondary reviewer with Dr. Sofia Serrano for ACL Rolling Review submissions.
- Evaluate research methodology, experimental design, and the validity of scientific claims.
Teaching Positions
Teaching Assistant
- Assist professors during class meetings and lead review sessions and debugging labs.
- Praised by students for clear explanations and care for their learning.
- See the teaching page for materials and sample activities.
Mentored Study Group Leader
- Lead two weekly review sessions covering course content for introductory CS.
- Create worksheets, slides, and practice problems to support student learning.
Publications & Preprints
- Eshbaugh, J., Tiwari, C., Silveyra, J.. “A Modular and Multimodal Generative AI Framework for Urban Building Energy Data: Generating Synthetic Homes.” Submitted to Energy & Buildings 2025. arXiv · DOI
- Eshbaugh, J.. “Fidelity Isn’t Accuracy: When Linearly Decodable Functions Fail to Match the Ground Truth.” 2025. arXiv · DOI
Talks & Posters
- “Generating Synthetic Homes: A Modular and Multimodal Generative AI Framework for Urban Building Energy Data.” Poster, Lafayette College Excel Scholars Poster Session, Easton, PA. Dec 2025.
- “Generating Synthetic Homes: A Modular and Multimodal Generative AI Framework for Urban Building Energy Data.” Poster, CyberAccelerate Poster Session at KINBERCON 2025, Lancaster, PA. Oct 2025.
- “Generating Synthetic Homes: A Modular and Multimodal Generative AI Framework for Urban Building Energy Data.” Poster, Lafayette College Bicentennial Weekend Poster Session, Easton, PA. Sep 2025.
Honors & Awards
- Marquis Scholarship, Lafayette College (2023).
- Dean’s List — F23, S24, F24, S25.
Skills
Programming: Python, Java, JavaScript, C, Standard ML · ML: PyTorch, TensorFlow, NumPy · Tools: Git, VS Code, JetBrains IDEs, LaTeX
Service
- Audio/Visual Specialist — Hope Alliance Church (Jul 2021–Present). Design weekly worship slides and sermon visuals; configure live stream and projection systems; troubleshoot AV issues; and reset the worship space after services.
- Audio/Visual Specialist — Lafayette DiscipleMakers Christian Fellowship (Sep 2023–Present). Run slides and AV for weekly meetings and events, supporting worship through technical service.
Certifications
- Deep Learning Specialization (Jun 2025).
Additional Information
Languages Spoken: English (native); French (fluent, professional competency).
Interests: Jazz vocals, trombone, piano, and composition; photography and photo editing; videography and video editing.